Search

Objective: The complexity of software systems, with their multifaceted functionalities and intricate source code structures, poses significant challenges for developers in identifying and resolving bugs. This study aims to address these challenges by proposing an efficient bug localization method that improves the accuracy and effectiveness of identifying faulty code based on bug reports. Method: We introduce a novel bug localization approach that integrates a Long Short-Term Memory (LSTM) attention mechanism with top-K code similarity learning. The proposed method preprocesses bug reports and source code files, calculates top-K code similarities using the BM25 algorithm, and trains an LSTM-Attention model to predict the most relevant buggy source code files. Results: The model was evaluated on six open-source projects (Tomcat, AspectJ, Birt, Eclipse, JDT, SWT) and demonstrated significant improvements over the baseline method, DNNLoc. Notably, the proposed approach improved accuracy across all projects, with average gains of 18% in prediction accuracy compared to the baseline. Conclusion: This study highlights the efficacy of combining similarity learning with attention mechanisms for bug localization. By streamlining debugging workflows and enhancing predictive accuracy, the proposed method offers a practical solution for improving software quality and reducing development costs. [ABSTRACT FROM AUTHOR]

The computational modeling of category learning is typically evaluated in terms of the model's accuracy. For a model to accurately infer category membership of stimuli, it has to have sufficient representational precision. Thus, many category learning models infer category representations that guide decision-making and the model's fitness is evaluated by its ability to accurately choose. Substantial decision-making research, however, indicates that noise plays an important role. Specifically, noisy representations are assumed to introduce an element of stochasticity to decision-making. Noise can be minimized at the cost of cognitive resource expenditure. Thus, a more biologically plausible model of category learning should balance representational precision with costs. Here, we tested an autoencoder model that learns categories (the six category structures introduced by Roger Shepard and colleagues) by balancing the minimization of error with minimization of resource usage. By incorporating the goal of reducing category complexity, the currently proposed model biases category decisions toward previously learned central tendencies. We show that this model is still able to account for category learning performance in a traditional category learning benchmark. The currently proposed model additionally makes some novel predictions about category learning that future studies can test empirically. The goal of this paper is to make progress toward development of an ecologically and neurobiologically plausible model of category learning that can guide future studies and theoretical frameworks. [ABSTRACT FROM AUTHOR]

In response to the limitations of traditional fuzzing approaches that rely on static mutators and fail to dynamically adjust their test case mutations for deeper testing, resulting in the inability to generate targeted inputs to trigger vulnerabilities, this paper proposes a directed fuzzing methodology termed DocFuzz, which is predicated on a feedback mechanism mutator. Initially, a sanitizer is used to target the source code of the tested program and stake in code blocks that may have vulnerabilities. After this, a taint tracking module is used to associate the target code block with the bytes in the test case, forming a high‐value byte set. Then, the reinforcement learning mutator of DocFuzz is used to mutate the high‐value byte set, generating well‐structured inputs that can cover the target code blocks. Finally, utilizing the feedback mechanism of DocFuzz, when the reinforcement learning mutator converges and ceases to optimize, the fuzzer is rebooted to continue mutating toward directions that are more likely to trigger vulnerabilities. Comparative experiments are conducted on multiple test sets, including LAVA‐M, and the experimental results demonstrate that the proposed DocFuzz methodology surpasses other fuzzing techniques, offering a more precise, rapid, and effective means of detecting vulnerabilities in source code. [ABSTRACT FROM AUTHOR]

This paper presents a method for lossless compression of images with fast decoding time and the option to select encoder parameters for individual image characteristics to increase compression efficiency. The data modeling stage was based on linear and nonlinear prediction, which was complemented by a simple block for removing the context-dependent constant component. The prediction was based on the Iterative Reweighted Least Squares (IRLS) method which allowed the minimization of mean absolute error. Two-stage compression was used to encode prediction errors: an adaptive Golomb and a binary arithmetic coding. High compression efficiency was achieved by using an author's context-switching algorithm, which allows several prediction models tailored to the individual characteristics of each image area. In addition, an analysis of the impact of individual encoder parameters on efficiency and encoding time was conducted, and the efficiency of the proposed solution was shown against competing solutions, showing a 9.1% improvement in the bit average of files for the entire test base compared to JPEG-LS. [ABSTRACT FROM AUTHOR]

This research, students' technology acceptance, KSB, community of inquiry (CoI), and social interaction space situations were examined in collaborative learning where Zoom was used as a CSCL tool in computer programming courses, and the structural relationships between these variables were revealed. The research was carried out on 162 university students studying in the field of computer science education and taking computer programming courses through flipped learning. Within the scope of the research, the students were divided into CSCL groups of three and carried out collaborative learning. The study was carried out according to the correlational research method, and path analysis was performed. Research findings for the CSCL activities carried out in the Zoom environment has been determined that technology acceptance, KSB, CoI, and social interaction space are at a high level. According to the results of the research, it has been seen that the technology acceptance status of the students in the CSCL environment has a direct and high impact on the KSB, CoI, and social interaction space. In addition, it was found that the increase in students' social interaction space perceptions in the CSCL process has a direct effect on CoI. [ABSTRACT FROM AUTHOR]

Recent advancements in intelligent tutoring systems (ITS) driven by artificial intelligence (AI) have attracted substantial research interest, particularly in the domain of computer programming education. Given the diversity in learners' backgrounds, cognitive abilities, and learning paces, the development of personalized tutoring strategies to support the effective attainment of learning objectives has become a critical challenge. This paper introduces personalized fuzzy logic strategies for intelligent programming tutoring (PerFuSIT), an innovative fuzzy logic-based module designed to select the most appropriate tutoring strategy from five available options, based on individual learner characteristics. The available strategies include revisiting previous content, progressing to the next topic, providing supplementary materials, assigning additional exercises, or advising the learner to take a break. PerFuSIT's decision-making process incorporates a range of learner-specific parameters, such as performance metrics, error types, indicators of carelessness, frequency of help requests, and the time required to complete tasks. Embedded within the traditional ITS framework, PerFuSIT introduces a sophisticated reasoning mechanism for dynamically determining the optimal instructional approach. Experimental evaluations demonstrate that PerFuSIT significantly enhances learner performance and improves the overall efficacy of interactions with the ITS. The findings highlight the potential of fuzzy logic to optimize adaptive tutoring strategies by customizing instruction to individual learners' strengths and weaknesses, thereby providing more effective and personalized educational support in programming instruction. [ABSTRACT FROM AUTHOR]

The quality of source code is negatively impacted by code smells. Since the term "code smell" originated, numerous attempts have been made to comprehend it by identifying it using various techniques, such as metric-based, heuristic-based, optimization-based, machine learning (ML)-based, etc. Among these, supervised machine learning (SML) has shown effectiveness in detecting code smells. However, SML techniques have significant limitations, including the dependency on expensive and high-quality labeled data, the need for representative training datasets, and the risk of introducing biases in labeled examples that lead to skewed predictions. To overcome these challenges, this study introduces a method that leverages unsupervised machine learning (UnML) along with feature engineering. Unlike SML, UnML does not require labeled data and minimizes potential biases. The proposed method was evaluated using four datasets containing different types of code smells and was compared with a previous study that used SML techniques. The results indicate that the UnML-based method is effective, achieving outcomes closely aligned with those from the SML approach. This method is especially beneficial in situations where labeled data is scarce or unavailable and can be used to identify new code smells, generate labeled data for SML and detect multiple code smells simultaneously within a codebase. [ABSTRACT FROM AUTHOR]

Programming has been considered a crucial "object to think with", which can integrate with various disciplines and facilitate learning in other areas. However, existing research primarily focuses on how children learn to code rather than how coding helps them to learn. To address this gap, a quasi-experimental case study was conducted with 31 kindergarten children, aiming to explore what young children could learn from a 4-week curriculum combining programmable robotics and road safety training. The findings indicate that teaching young children to code not only develops their computational thinking (CT), but also enhances their road safety knowledge and skills. The results contribute to a more comprehensive understanding of early childhood coding education and may inspire the innovation and diversification of coding curricula for young children. [ABSTRACT FROM AUTHOR]

Gender balance in computing education is a decades-old issue that has been the focus of much previous research. In K-12, the introduction of mandatory computing education goes some way to giving all learners the opportunity to engage with computing throughout school, but a gender imbalance still persists when computer science becomes an elective subject. The study described in this paper investigates whether introducing pair programming would make a difference to primaryaged girls' attitudes to computing and intent to study the subject in the future. A randomised controlled trial (RCT) was designed and implemented around a 12-week intervention with 785 female pupils between the ages of 8 and 10 years, alongside a qualitative evaluation investigating teachers' and pupils' experience of the interventions and the development of materials and teacher preparation resources. The results of the RCT showed no statistically significant changes in student attitudes towards computing or intent to study further, although the qualitative data indicated that both teachers and pupils found the interventions engaging and enjoyable. Themes emerging from the qualitative data point to the importance of collaboration in supporting a development in pupil confidence. Overall, these results emphasise the societal and systemic barriers around computer science and technology engagement across genders that persist despite many initiatives being implemented over many years. [ABSTRACT FROM AUTHOR]

To solve the problems of insufficient ability of current deep hashing algorithms to extract cross-scale features and difficulty in fitting the global similarity distribution of data, this paper proposed a deep hashing method based on cross-scale Vision Transformer. Firstly, the method utilized pyramid convolution and cross-scale attention mechanism to construct a multi-level encoder to capture the rich semantic information of the image. Secondly, the method proposed a proxy based deep hashing algorithm. This algorithm generated hash proxies for each category, allowing hash codes to learn discriminative class features to reduce the distance from hash proxies of the same category and fit the global similarity distribution of the data. Finally, the method added an angle margin term between the hash proxy and the hash code to expand intra class similarity and inter class differences to generate hash codes with high discriminability. The experimental results conducted on CIFAR-10, ImageNet-100, NUS Wide, and MS COCO show that the average retrieval accuracy of the algorithm is 4.42%, 19.61%, 0.35%, and 15.03% higher than the suboptimal method, respectively, demonstrating the effectiveness of the algorithm. [ABSTRACT FROM AUTHOR]

Almost all educational systems still rely on traditional teaching methods: children must absorb huge uncorrelated amounts of information from various disciplines, multidisciplinary aspects are not developed, the fun factor is missing in most of the classes. Many children sometimes do not fully understand the presented theoretical concepts and their applicability in real life, so they lose interest and end up learning in a mechanical way. DigiMathArt is a teaching method designed to present Mathematics, Programming, and 3D modelling concepts in a multidisciplinary approach, in an attractive manner, combining theory and practice in a game-like 3D virtual environment. Its main principles are task-divided project-based approach, practicality, pre-theoretical learning, multidisciplinarity, 3D visualization. The method was used to build an ample content of lessons and exercises, covering a large part of the subjects corresponding to K-12 education, and its strength was proven through a pilot involving 100 children. Their positive feedback highlights the attractiveness of the method and its ability to teach theoretical concepts through practice. [ABSTRACT FROM AUTHOR]

Visual storytelling, particularly through dance choreographies as showcased in previous AlgoRythmics performances, has been effective in communicating relatively straightforward algorithms in an engaging and memorable way. Nevertheless, when addressing complex algorithmic concepts, an approach with greater expressiveness and flexibility becomes necessary. Consequently, this study introduces stage performances as an innovative solution, using cinematic representation to successfully convey and communicate these intricate concepts and processes. To evaluate the effectiveness of this approach, a short film was designed, produced, and showcased to a second-semester CS2 university course audience studying programming techniques. Following an opening scene that establishes the context, the subsequent three acts vividly depict ad hoc, greedy, and dynamic programming solutions in response to the posed programming challenge. After the screening, a questionnaire was administered, built on four key constructs of the Technology Acceptance Model, as well as other potential facilitating factors. The study reveals 100% positive perceptions of educational benefits, with the vast majority of students expressing agreement regarding the utility, enjoyment, engagement, creativity, filmic quality, and cognitive benefits of short films. Additionally, a remarkable 96% reported the intent to utilize this approach. Our subsequent Structural Equation Modeling analysis discovered that students whose learning styles were in sync with this approach demonstrated a robust correlation between their perception of the method's value, their enjoyment of the process, and their overall attitude towards this pedagogical method. This study confirms the potential of visual storytelling through short films as an effective tool for delivering programming education. The findings provide valuable insights for computer science educators seeking to engage learners and convey complex information in an attractive and effective way. [ABSTRACT FROM AUTHOR]

This study investigated the impact of problematization-oriented scaffolding and structuring-oriented scaffolding, incorporated within instructional videos, on students' computational thinking and their performance in programming education. We recruited 86 participants from three senior classes at a high school. Each of the three classes was assigned to one of the experimental conditions: the problematization-oriented scaffolding group, the structuring-oriented scaffolding group, and the control group, which received no scaffolding. Pre- and post-tests were conducted to assess students' computational thinking and programming performance during a three-week period. We observed significant improvements in the computational thinking skills of the problematization-oriented scaffolding group, particularly in the algorithmic thinking and cooperativity subdimensions, when compared to the control group. Structuring-oriented scaffoldings in videos also helped enhance students' computational thinking, especially in the algorithmic thinking subdimension. In contrast, students in the control group primarily improved their problem-solving skills, although the difference across the three groups was not significant. Furthermore, this study revealed that the use of learning scaffolds, whether problematization-oriented or structured, significantly contributes to students' learning achievements in comparison to the control group. These findings emphasize the importance of selecting the appropriate scaffolding approach to enhance specific dimensions of computational thinking. [ABSTRACT FROM AUTHOR]

High-level network management is achieved by using a software-defined network (SDN). The SDN clearly differentiates the data plane using effective programs. The centralized controllers in the SDN have much vulnerability to various types of attacks by fake hosts. One of the ways to attack is link coding. This link coding attack easily disconnects the data plane from the control plane by creating more traffic in the network. This makes SDN poor centralized manageability. In this paper, we demonstrate the vulnerability of the SDN control layer to link coding and how the attack strategy differs when targeting traditional networks which primarily involves attacking the links directly. In link coding, the attacker employs bots to surreptitiously send low-rate legitimate traffic on the control channel which ultimately results in disconnecting the control plane from the data plane. To address this challenge, a machine learning-based classifier is used to alleviate link coding in SDN. This paper proposed a Random Forest-based Adaptive Gradient Algorithm (RF-AGA). The gradient descent optimization method uses standard gradient as the objective function to search the target but adaptive gradient uses adaptive sizes for each input data in objective functions. Here deep learning techniques are used to classify the network traffics and is implemented as an extension module in the Floodlight controller. The accuracy rate of Naïve Bayesian was 81.11%, J-48 87.54%, Random Forest 91.02%, and the proposed work got 93.15%. [ABSTRACT FROM AUTHOR]

Perception and action rely on integrating or binding different features of stimuli and responses. Such bindings are short-lived, but they can be retrieved for a limited amount of time if any of their features is reactivated. This is particularly true for stimulus–response bindings, allowing for flexible recycling of previous action plans. A relation to learning of stimulus–response associations suggests itself, and previous accounts have proposed binding as an initial step of forging associations in long-term memory. The evidence for this claim is surprisingly mixed, however. Here we propose a framework that explains previous failures to detect meaningful relations of binding and learning by highlighting the joint contribution of three variables: (a) decay, (b) the number of repetitions, and (c) the time elapsing between repetitions. Accounting for the interplay of these variables provides a promising blueprint for innovative experimental designs that bridge the gap between immediate bindings on the one hand and lasting associations in memory on the other hand. [ABSTRACT FROM AUTHOR]

While research has shown that students' epistemic beliefs can be a strong predictor of their academic performance, cognitive abilities, or self-efficacy, studies of this topic in computer education are rare. The purpose of this study was twofold. First, it aimed to validate a newly developed questionnaire for measuring students' epistemic beliefs about computer programming at middle school level. Second, the study aimed to understand how students' epistemic beliefs predict their computational thinking disposition and computer programming self-efficacy. A total of 406 middle school students completed a survey consisting of three research instruments: the Computer Programming Epistemic Belief Inventory (CPEBI), the Computational Thinking Scale (CTS), and the Computer Programming Self-Efficacy Scale (CPSES). The results indicated that the epistemic belief of "uncertainty" was a significant and positive predictor of the "abstraction" and "generalization" dimensions of computational thinking dispositions. In addition, the belief of "complexity" was an antecedent to three dimensions of computational thinking, namely, "algorithmic thinking," "evaluation," and "generalization." Only the "evaluation" dimension of CT disposition had significant and positive relationships with "programming self-efficacy." The roles of "uncertainty" and "complexity" of computer programming epistemic beliefs are discussed. Implications and suggestions for future computer programming education, especially for young students, are provided. [ABSTRACT FROM AUTHOR]

This paper examines healthcare governance in Sweden, focusing on two key trends: "Knowledge-based management" and "Trust-based management" (in Swedish, "kunskapsstyrning" and "tillitsbaserad styrning"). These trends are analysed through three dimensions of governance, rooted in theories of knowledge and learning, each represented on a continuum. The first dimension assesses the extent to which governance prioritizes tacit versus explicit knowledge. The second considers the locus of knowledge, distinguishing whether it resides within individuals or social groups, thereby influencing whether learning is oriented toward individual or collective processes. The third dimension contrasts learning that reproduces established knowledge with learning that generates innovation and new knowledge. Through detailed document analysis, the paper highlights disparities between Knowledge-based management and Trust-based management at the policy level, particularly in their discursive constructions of knowledge and learning across these three dimensions. The analysis suggests that these governance approaches sometimes conflict, requiring a choice between pursuing the ideals of Knowledge-based management or Trust-based management, while at other times, they can coexist side by side. This paper contributes to the discourse on healthcare governance by providing an analytical framework for understanding how governance approaches embody distinct "codes of knowing" and support varying modes of knowledge and learning processes. [ABSTRACT FROM AUTHOR]

In the video compression industry, video compression tailored to machine vision tasks has recently emerged as a critical area of focus. Given the unique characteristics of machine vision, the current practice of directly employing conventional codecs reveals inefficiency, which requires compressing unnecessary regions. In this paper, we propose a framework that more aptly encodes video regions distinguished by machine vision to enhance coding efficiency. For that, the proposed framework consists of deep learning-based adaptive switch networks that guide the efficient coding tool for video encoding. Through the experiments, it is demonstrated that the proposed framework has superiority over the latest standardization project, video coding for machine benchmark, which achieves a Bjontegaard delta (BD)-rate gain of 5.91% on average and reaches up to a 19.51% BD-rate gain. [ABSTRACT FROM AUTHOR]

Odour processing exhibits multiple parallels between vertebrate and invertebrate olfactory systems. Insects, in particular, have emerged as relevant models for olfactory studies because of the tractability of their olfactory circuits. Here, we used fast calcium imaging to track the activity of projection neurons in the honey bee antennal lobe (AL) during olfactory stimulation at high temporal resolution. We observed a heterogeneity of response profiles and an abundance of inhibitory activities, resulting in various response latencies and stimulus-specific post-odour neural signatures. Recorded calcium signals were fed to a mushroom body (MB) model constructed implementing the fundamental features of connectivity between olfactory projection neurons, Kenyon cells (KC), and MB output neurons (MBON). The model accounts for the increase of odorant discrimination in the MB compared to the AL and reveals the recruitment of two distinct KC populations that represent odorants and their aftersmell as two separate but temporally coherent neural objects. Finally, we showed that the learning-induced modulation of KC-to-MBON synapses can explain both the variations in associative learning scores across different conditioning protocols used in bees and the bees' response latency. Thus, it provides a simple explanation of how the time contingency between the stimulus and the reward can be encoded without the need for time tracking. This study broadens our understanding of olfactory coding and learning in honey bees. It demonstrates that a model based on simple MB connectivity rules and fed with real physiological data can explain fundamental aspects of odour processing and associative learning. [ABSTRACT FROM AUTHOR]

In the twenty-first century, programming courses have become integral to primary school education worldwide, emphasizing the importance of computational thinking. However, research on computational thinking within programming courses has primarily focused on text-based programming languages, particularly at the secondary and higher education levels. There has been a lack of research on the influence of block-based programming languages on computational thinking in primary school. To fill this research gap, our study investigated the impact of the flipped classroom approach combined with block-based programming tools on developing computational thinking in primary school students in Macao. Over nine weeks, 20 third-grade students engaged with mBlock and Codey Rocky programmable hardware, with pretest and posttest assessments using the Programming Computational Thinking Scale (PCTS). Results indicated significant improvements in Computational Concepts and Computational Practices, though changes in Computational Perspectives were not statistically significant. Moreover, the Pearson correlation tests showed strong positive correlations among the dimensions of computational thinking and programming achievement. The findings demonstrated the flipped classroom approach, which emphasizing student-centered learning and active participation, effectively enhanced computational thinking skills. This study provided valuable insights for educators and curriculum developers aiming to improve programming education and computational thinking in primary schools, highlighting the potential of innovative teaching methods to meet twenty-first century educational needs. Using block-based programming tools combined with the flipped classroom approach offers a promising avenue for developing comprehensive computational thinking skills in young learners, ensuring they are better prepared for future technological challenges. [ABSTRACT FROM AUTHOR]

College and university remain the dominant academic pathways for graduating Canadian high school students. However, apprenticeships can provide alternative education pathways for a significant proportion of the population. Despite the push to undergraduate programming, some students end up in fields that don't suit their interests. To pursue academic programming that better suits personal interests and academic aptitudes, students have the option of transferring schools. However, no study to-date has explored the transfer reasons and process of students moving between university and apprenticeships in Canada. To address this knowledge gap, we collected qualitative interview data from post-secondary administrators who work directly with transfer students. We also collected interview data from Ontario-based students who transferred to apprenticeships from university. Analysis of our interview data identifies mechanisms that influenced participants' education pathway choices. [ABSTRACT FROM AUTHOR]

Over the past 20 years, there has been a growing interest in the processing of tool objects, and in the spatial S-R correspondence effects obtained with pictures of manipulable objects. Beyond the original affordance activation account, a location coding account has been proposed for such behavioral effects, which states that the location of the visually salient portion of an object speeds up spatially aligned motor responses and slows down misaligned ones. Furthermore, an additional action coding account has been proposed, according to which it is the direction of the action of the perceived object (e.g., "pouring tea leftward" when the spout of a teapot is leftward-oriented) that biases motor performance. We investigated this action coding account of S-R correspondence effects by comparing the reaction time (RT) performance of two groups: younger (6 to 9 years old) and older (10 to 13 years old) children. We assumed that knowledge of tool objects and the activation of action codes from object perception is proportional to age. Therefore, a larger correspondence effect was hypothesized for the older relative to the younger children's group. Consistently, a 34 ms correspondence effect was observed in the older children relative to the younger ones (18 ms). The results support the view that action direction is a constitutive feature of tool objects, which is subject to experience, and thus, to increasing knowledge over time. [ABSTRACT FROM AUTHOR]

Constructed response items that require the student to give more detailed and elaborate responses are widely applied in large-scale assessments. However, the hand-craft scoring with a rubric for massive responses is labor-intensive and impractical due to rater subjectivity and answer variability. The automatic response coding method, such as the automatic scoring of short answers, has become a critical component of the learning and assessment system. In this paper, we propose an interactive coding system called ASSIST to efficiently score student responses with expert knowledge and then generate an automatic score classifier. First, the ungraded responses are clustered to generate specific codes, representative responses, and indicator words. The constraint set based on feedback from experts is taken as training data in metric learning to compensate for machine bias. Meanwhile, the classifier from responses to code is trained according to the clustering results. Second, the experts review each coded cluster with the representative responses and indicator words to score a rating. The coded cluster and score pairs will be validated to ensure inter-rater reliability. Finally, the classifier is available for scoring a new response with out-of-distribution detection, which is based on the similarity between response representation and class proxy, i.e., the weight of class in the last linear layer of the classifier. The originality of the system developed stems from the interactive response clustering procedure, which involves expert feedback and an adaptive automatic classifier that can identify new response classes. The proposed system is evaluated on our real-world assessment dataset. The results of the experiments demonstrate the effectiveness of the proposed system in saving human effort and improving scoring performance. The average improvements in clustering quality and scoring accuracy are 14.48% and 18.94%, respectively. Additionally, we reported the inter-rater reliability, out-of-distribution rate, and cluster statistics, before and after interaction. [ABSTRACT FROM AUTHOR]

Code models have made significant advancements in code intelligence by encoding knowledge about programming languages. While previous studies have explored the capabilities of these models in learning code syntax, there has been limited investigation on their ability to understand code semantics. Additionally, existing analyses assume that the number of edges between nodes at the abstract syntax tree (AST) is related to syntax distance, and also often require transforming the high-dimensional space of deep learning models to a low-dimensional one, which may introduce inaccuracies. To study how code models represent code syntax and semantics, we conduct a comprehensive analysis of seven code models, including four representative code pre-trained models (CodeBERT, GraphCodeBERT, CodeT5, and UnixCoder) and three large language models (LLMs) (StarCoder, CodeLlama and CodeT5+). We design four probing tasks to assess the models' capacities in learning both code syntax and semantics. These probing tasks reconstruct code syntax and semantics structures (AST, control dependence graph (CDG), data dependence graph (DDG), and control flow graph (CFG)) in the representation space. These structures are core concepts for code understanding. We also investigate the syntax token role in each token representation and the long dependency between the code tokens. Additionally, we analyze the distribution of attention weights related to code semantic structures. Through extensive analysis, our findings highlight the strengths and limitations of different code models in learning code syntax and semantics. The results demonstrate that these models excel in learning code syntax, successfully capturing the syntax relationships between tokens and the syntax roles of individual tokens. However, their performance in encoding code semantics varies. CodeT5 and CodeBERT demonstrate proficiency in capturing control and data dependencies, whereas UnixCoder shows weaker performance in this aspect. We do not observe LLMs generally performing much better than pre-trained models. The shallow layers of LLMs perform better than their deep layers. The investigation of attention weights reveals that different attention heads play distinct roles in encoding code semantics. Our research findings emphasize the need for further enhancements in code models to better learn code semantics. This study contributes to the understanding of code models' abilities in syntax and semantics analysis. Our findings provide guidance for future improvements in code models, facilitating their effective application in various code-related tasks. [ABSTRACT FROM AUTHOR]

Metamorphic testing (MT) is an effective testing technique having a broad range of applications. One key task for MT is the identification of metamorphic relations (MRs), which is a fundamental mechanism in MT and is critical to the automation of MT. Prior studies have proposed approaches for predicting MRs (PMR). One major idea behind these PMR approaches is to represent program source code information via manually designed code features and then to apply machine‐learning–based classifiers to automatically predict whether a specific MR can be applied on the target program. Nevertheless, the human‐involved procedure of selecting and extracting code features is costly, and it may not be easy to obtain sufficiently comprehensive features for representing source code. To overcome this limitation, in this study, we explore and evaluate the effectiveness of code representation learning techniques for PMR. By applying neural code representation models for automatically mapping program source code to code vectors, the PMR procedure can be boosted with learned code representations. We develop 32 PMR instances by, respectively, combining 8 code representation models with 4 typical classification models and conduct an extensive empirical study to investigate the effectiveness of code representation learning techniques in the context of MR prediction. Our findings reveal that code representation learning can positively contribute to the prediction of MRs and provide insights into the practical usage of code representation models in the context of MR prediction. Our findings could help researchers and practitioners to gain a deeper understanding of the strength of code representation learning for PMR and, hence, pave the way for future research in deriving or extracting MRs from program source code. [ABSTRACT FROM AUTHOR]

The significance of teachers' self-efficacy in coding education within K-12 settings has grown substantially. However, the literature lacks subject-specific measurement tools tailored to assess teachers' self-efficacy in coding instruction. This study adopted a mixed methods approach to develop a reliable and valid instrument for measuring computer science (CS) teachers' self-efficacy in teaching coding. The scale development involved a rigorous process encompassing item generation, expert validation, and pilot testing. Importantly, this process unfolded in seven steps with two distinct phases, and each phase involved independent sample groups. Subsequently, a comprehensive survey was administered to two samples of CS teachers (n = 318, n = 295) to assess the scale's psychometric properties. The results revealed robust internal consistency and construct validity of the 20-item Coding Teaching Self-Efficacy Scale (CTSES) with four intercorrelated dimensions: student motivation, subject knowledge, classroom management, and material development. Furthermore, additional analyses revealed the significant impact of teaching experience and self-reported coding knowledge level on teachers' self-efficacy in teaching coding. The implications of this study hold significance for both practitioners and researchers to understand teachers' self-efficacy in teaching coding and to explore its relation to teacher training, curriculum development, and the broader advancement of coding education within school settings. [ABSTRACT FROM AUTHOR]

The COVID-19 pandemic has forced a significant increase in the utilization of video-based e-learning platforms for programming education. These platforms never considered the essential attributes of student characteristics and learning preferences while designing such a problematic subject having high dropout and failure rates. The traditional e-learning environments deliver instructional videos to the learners by assuming all learners have a single learning preference. Moreover, existing learning style models need to address the recent requirements of e-learning paradigms. To address this issue, this paper presents a novel learning style model tailored for instructional video-based programming e-learning environments that map individual learning preferences with various video design patterns. An adaptive e-learning environment was employed to assess the effectiveness of the proposed model that leveraged a decision tree classifier to divide learners into four preferences. In a paired experimental design, 195 first-year undergraduate students were randomly assigned to one of three groups where learner scores and feedback were taken as evaluation metrics. The control group partook without instructional videos for the entire semester of six months. During the same period, experimental group-1 learned with a traditional video-based e-learning environment, and experimental group-2, with the proposed learning style model, enabled an adaptive e-learning environment. Based on the proposed decision tree learning model, it is understood that the intervention group showed significant improvements in knowledge acquisition, grade, and positive feedback compared to the other groups. Hence, the proposed model is highly recommended for traditional programming e-learning environments to deliver instructional videos based on learners' learning preferences. [ABSTRACT FROM AUTHOR]

As a highly promising technique in the field of similarity search, the hashing-based image retrieval algorithm has received continued attention in recent years because of its strong ability to efficiently provide accurate results when measuring similarities between data instances in the binary space. At the level of practical applications, it is necessary to consider both the optimization of the hashing algorithm itself and the mining of image semantic features to produce high-quality hash codes with a strong semantic resolution, enabling the performance maximisation of such a neighbour search system on complex scenario images. To this end, a unified deep hashing framework termed Dual Enhanced Semantic Hashing (DESH) is proposed in this work. Specifically, it benefits the fast image retrieval in two aspects: 1) By taking advantage of dynamic multi-scale fusion and graph encoding networks, the dual enhanced feature learning significantly strengthens the semantic feature representation by jointly exploring and encoding the local multi-scale information with the high-order adjacency relationship between original images; 2) With the joint optimization of diverse loss functions, the binary semantic modelling process seamlessly module the image semantic information within the hash function learning in the code generation, aiming to generate discriminative hash codes to refine the retrieval performance eventually. By conducting extensive experiments on public datasets, the retrieval results further validate the claims of the proposed DESH by exhibiting its superior performance against competitive state-of-the-art baselines. [ABSTRACT FROM AUTHOR]

AbstractIn recent years, the integration of programming education into school curricula has reemerged, acknowledging its importance and potential to enhance critical cognitive abilities among students. This systematic review delves into the influence of programming education on higher-order executive functions (EFs), including problem-solving, reasoning, and planning, in school-aged children. The systematic literature search identified 18 studies meeting inclusion criteria. Of these, 14 studies showed significant positive effects of coding education on EFs, particularly in problem-solving and planning. However, limitations such as the absence of randomized control trials and age-related clustering of skills underscore the need for more robust research to ascertain the impact of coding education on EFs. [ABSTRACT FROM AUTHOR]

Code clone detection is a critical problem in software development and maintenance domains. It aims to identify functionally identical or similar code fragments within an application. Existing works formulate the code clone detection task as a binary classification problem which predicts a code pair as a clone or not based on a pre‐defined threshold. In reality, there are various types of code clone subject to the degree of how a pair of code fragments are similar to each other. To investigate the effect of different code clone detection manners on the clone detection result, we propose Gated Recurrent Residual Learning Networks (GRRLN), a novel neural network model for code clone detection. To train GRRLN, we first represent each code fragment as a statement‐level tree sequence derived from the whole abstract syntax tree (AST). Then, a gated recurrent neural network with residual connections is adopted to fully extract the semantics of all individual statement trees together with their dependency relationships across the input statement sequence. Finally, the output representations of code fragments by GRRLN are used for similarity calculation and clone detection. We evaluate GRRLN using two real‐world datasets for code clone detection and clone type classification. Experiments show that GRRLN achieves promising and compelling results and meanwhile needs significantly less time and memory consumption compared with the state‐of‐the‐art methods. [ABSTRACT FROM AUTHOR]

The pedagogical of computer programming education is being enriched and improved through the interactive learning material. Visualization, modeling, and internet platforms for developing interactive visual skills are only a few examples of the types of specialized learning material currently accessible for use in a wide range of computing classes. There are some specific challenges related to the implementation of active learning, such as insufficient time for class, an increase in preparation, implementing students' engagement in extensive courses, and a lack of necessary materials, technology, or supplies. Computer vision is a subfield of AI that allows machines to learn from visual data (such as photos, movies, and other digital media) and then act on or offer solutions to problems. To enhance the efficiency of intelligent interactive learning and practice, this article incorporates a visual machine vision analytical framework under the guidance of Artificial intelligence to create a Machine-Vision-based Smart Education Assistance System (MV-SEAS). Visualization speeds up and simplifies regular communication by consolidating several forms of information into a single visual representation. This study discusses how visualizing information is crucial for students' initial knowledge acquisition and continued education and development. The seamless amalgamation of automated innovative education analyses and interactive visualizations is emphasized. The paper aimed to identify and characterize the technical challenges mentioned above must be surmounted to make it simpler for computer educators to discover, adopt, and tailor intelligent learning materials. The study concludes by proposing an MV-SEAS for storing, integrating, and disseminating smart educational data. It investigates whether it can be done using existing standards and guidelines. In the end, this essay combines trials to prove the effectiveness of the proposed smart education method. The findings demonstrate that interactive visualization of AI-assisted smart education may effectively combine subject experts' information with educators' experience to produce more powerful and easily intelligible machine intelligence. [ABSTRACT FROM AUTHOR]

In the face of the challenges posed by the COVID-19 pandemic, the hybrid teaching model has garnered significant attention for its combination of the depth of traditional education with the convenience of distance learning. Focusing on the domain of computer programming language instruction, this study innovatively designs a hybrid teaching strategy aimed at fully exploiting the flexibility of its teaching design and the variety of pedagogical approaches. The strategy integrates face-to-face teaching with online autonomous learning, incorporating project-based teaching methodologies and immediate feedback mechanisms to facilitate active student engagement and deep learning. Through a year-long practice in a C++ programming course, encompassing 68 students, the study empirically validates the effectiveness of the hybrid teaching approach. It not only demonstrates remarkable educational outcomes, enhancing the quality of programming instruction and student satisfaction with their learning experience, but also employs Bayesian analysis to delve into the relationship between learning trajectories and students' sense of self-efficacy. By focusing on key indicators during the learning process, such as the timeliness and quality of online learning, laboratory work, and project assignments, the study then utilizes Bayesian models to directly assess the impact of these learning behavior metrics on students' perceived self-efficacy. The findings reveal that students with outstanding academic achievements exhibit higher levels of self-efficacy, confirming that academic performance can reasonably reflect teaching effectiveness and provide a quantifiable basis for assessing individual learning progress. Consequently, this research not only contributes a novel strategy to computer programming education practice but also offers a valuable reference for the application of hybrid teaching models in other disciplines. Furthermore, it promotes indepth contemplation on post-pandemic innovations in teaching modes and issues of educational equity, laying a solid foundation for constructing a more adaptive and inclusive future education system. [ABSTRACT FROM AUTHOR]

Recently, there has been a significant increase in interest in cross-modal hashing technology. For hash code learning, most previous supervision methods use label information to create a similarity matrix in a straightforward manner. However, there are still the following challenges: (1) The asymmetric similarity matrix method only considers the similarity of labels, and the discriminant constraint in Hamming space is ignored; (2) there are optimization errors between the cross-modal semantic correlation of the Hamming space and the nonlinearity of the feature space; and (3) the cross-modal hash matrix is in a dynamic state during the optimization process, and the hash code is easily disturbed and there is bit uncertainty. To this end, we propose the Discriminative Latent Semantics-preserving Similarity Embedding Cross-modal Hashing (DLSSECH) method. Specifically, to reduce the quantization error, we introduce a non-asymmetric similarity decomposition based on orthogonal rotation bias. It can decompose the bitwise correlation of the learned hash code to capture more discriminative semantic information and compact hash code. In addition, to capture the cross-modal semantic correlation of nonlinear feature transformation and reduce quantization error, we propose the latent correlation error matrices based on orthogonal rotation decomposition. The model maintains the maximum difference in semantic dependencies between the projected data and its semantic representation. Finally, we use a sparse common hash matrix and non-asymmetric similarity decomposition factors to solve the uncertainty of the dynamic changes of the hash code. The effectiveness of DLSSECH has demonstrated through experiments on four cross-modal retrieval datasets, where it outperformed some state-of-the-art methods. [ABSTRACT FROM AUTHOR]

Coding education has become a compulsory course for cultivating children's computational thinking, and its tools in a gamification context can further enhance children's learning enthusiasm. Through the literature review, the researchers identified gamification, unplugging, and low cost as important design requirements for children's coding education tools. In order to verify the effectiveness of teaching design, this study specially developed a gamified coding education tool, "Coding Adventure," which allows child learners to take on gamified roles in real-life scenarios and use instruction cards to complete tasks. While testing the prototype in the on-site teaching environment, the researchers invited 12 kindergarten and elementary school teachers to observe the testing process. Then, they formed a focus group discussion to obtain their feedback on the education tool using MAXQDA software for qualitative analysis. According to the results, teacher participants generally recognize the design concepts of gamification, unplugging, and low-cost, and believe that role-playing and rewards in gamification, as well as the use of safe and simple materials to match the existing teaching environment of the school, are successful design innovations. In addition, gamified storytelling has also been proven to enhance students' team communication using this prototype. Overall, this study proves the effectiveness of the design concepts of gamification, unplugged and low cost on the perceived usefulness and ease of use of the coding education tool system under the Technology Acceptance Model theory framework. These studentcentered design concepts will provide valuable experience for the further development of unplugged coding education tools. [ABSTRACT FROM AUTHOR]

Prior studies have demonstrated that deep learning-based approaches can enhance the performance of source code vulnerability detection by training neural networks to learn vulnerability patterns in code representations. However, due to limitations in code representation and neural network design, the validity and practicality of the model still need to be improved. Additionally, due to differences in programming languages, most methods lack cross-language detection generality. To address these issues, in this paper, we analyze the shortcomings of previous code representations and neural networks. We propose a novel hierarchical code representation that combines Concrete Syntax Trees (CST) with Program Dependence Graphs (PDG). Furthermore, we introduce a Tree-Graph-Gated- Attention (TGGA) network based on gated recurrent units and attention mechanisms to build a Hierarchical Code Representation learning-based Vulnerability Detection (HCRVD) system. This system enables cross-language vulnerability detection at the function-level. The experiments show that HCRVD surpasses many competitors in vulnerability detection capabilities. It benefits from the hierarchical code representation learning method, and outperforms baseline in cross-language vulnerability detection by 9.772% and 11.819% in the C/C++ and Java datasets, respectively. Moreover, HCRVD has certain ability to detect vulnerabilities in unknown programming languages and is useful in real open-source projects. HCRVD shows good validity, generality and practicality. [ABSTRACT FROM AUTHOR]

Copyright of Instructional Technology & Lifelong Learning is the property of Instructional Technology & Lifelong Learning and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

Copyright of Journal of University Research / Üniversite Araştırmaları Dergisi is the property of Journal of University Research and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

Summarizing source code by natural language aims to help developers better understand existing code, making software development more efficient. Since source code is highly structured, recent research uses code structure information like Abstract Semantic Tree (AST) to enhance the structure understanding rather than a normal translation task. However, AST can only represent the syntactic relationship of code snippets, which can not reflect high-level relationships like control and data dependency in the program dependency graph. Moreover, researchers treat the AST as the unique structure information of one code snippet corresponding to one summarization. It will be easily affected by simple perturbations as it lacks the understanding of code with similar structure. To handle the above problems, we build CogCol, a Code graph-based Contrastive learning model. CogCol is a Transformer-based model that converts code graphs into unique sequences to enhance the model's structure learning. In detail, CogCol uses supervised contrastive learning by building several kinds of code graphs as positive samples to enhance the structural representation of code snippets and generalizability. Moreover, experiments on the widely used open-source dataset show that CogCol can significantly improve the state-of-the-art code summarization models under Meteor, BLEU, and ROUGE. [ABSTRACT FROM AUTHOR]

Code search, which locates code snippets in large code repositories based on natural language queries entered by developers, has become increasingly popular in the software development process. It has the potential to improve the efficiency of software developers. Recent studies have demonstrated the effectiveness of using deep learning techniques to represent queries and codes accurately for code search. In specific, pre-trained models of programming languages have recently achieved significant progress in code searching. However, we argue that aligning programming and natural languages are crucial as there are two different modalities. Existing pre-train models based approaches for code search do not effectively consider implicit alignments of representations across modalities (inter-modal representation). Moreover, the existing methods do not take into account the consistency constraint of intra-modal representations, making the model ineffective. As a result, we propose a novel code search method that optimizes both intra-modal and inter-modal representation learning. The alignment of the representation between the two modalities is achieved by introducing contrastive learning. Furthermore, the consistency of intra-modal feature representation is constrained by KL-divergence. Our experimental results confirm the model's effectiveness on seven different test datasets. This paper proposes a code search method that significantly improves existing methods. Our source code is publicly available on GitHub.1 [ABSTRACT FROM AUTHOR]

Today computing is intricate in all aspects of our lives, beginning with communications and education to banking, information security, health, shopping, and social media. Development of the computing is proportional to the development of software which is becoming a serious part of all daily lives. This paper, therefore, assessed the impact of teaching Computer Programming Language in Tanzanian primary schools. To achieve this, a new tool for teaching and learning a computer programming language was developed. Furthermore, a two-tier approach was adopted where the training started on an algorithm first, then if, loops and case conditionals as the third concept inside the first tier. On the other hand, variables, lists, and functions are hooked on the second tier. Results indicate that students acquired together reasoning learning in a conceptual argument of view and skill performance. Students were able to accomplish the expertise to practice or apply the theories of programming knowledge in programming-connected tasks. [ABSTRACT FROM AUTHOR]

The integration of the Arduino board into educational settings has penetrated across various educational levels. The teaching of this subject can be accomplished by (a) using real components in breadboards, (b) prefabricated modular boards that snap together, and (c) utilizing computer simulations. Yet, it is unknown which interface offers a more effective learning experience. Therefore, this experimental study aims to compare the effectiveness of these interfaces in a series of three laboratory exercises involving 110 university students, who were divided into three groups: (a) the first group used a tangible user interface, implementing circuits on breadboards, (b) the second group also used a tangible interface but with modular boards, and (c) the third group used a graphical user interface to simulate circuits using Tinkercad. For each laboratory exercise, students completed both pretests and posttests. Also, they provided feedback through five Likert-type attitude questions regarding their experiences. In terms of data analysis, t-tests, ANOVA, and ANCOVA, along with bootstrapping, and principal component analysis were employed. The results suggest that among the participants, those who used a graphical user interface stated that their understanding of the interconnection of components in microcontroller circuits was enhanced, while students with previous experience in microcontroller labs found the circuit creation process easier than students without experience. [ABSTRACT FROM AUTHOR]

This article studies deep unsupervised hashing, which has attracted increasing attention in large-scale image retrieval. The majority of recent approaches usually reconstruct semantic similarity information, which then guides the hash code learning. However, they still fail to achieve satisfactory performance in reality for two reasons. On the one hand, without accurate supervised information, these methods usually fail to produce independent and robust hash codes with semantics information well preserved, which may hinder effective image retrieval. On the other hand, due to discrete constraints, how to effectively optimize the hashing network in an end-to-end manner with small quantization errors remains a problem. To address these difficulties, we propose a novel unsupervised hashing method called HARR to learn discriminative and high-quality hash codes. To comprehensively explore semantic similarity structure, HARR adopts the Winner-Take-All hash to model the similarity structure. Then similarity-preserving hash codes are learned under the reliable guidance of the reconstructed similarity structure. Additionally, we improve the quality of hash codes by a bit correlation reduction module, which forces the cross-correlation matrix between a batch of hash codes under different augmentations to approach the identity matrix. In this way, the generated hash bits are expected to be invariant to disturbances with minimal redundancy, which can be further interpreted as an instantiation of the information bottleneck principle. Finally, for effective hashing network training, we minimize the cosine distances between real-value network outputs and their binary codes for small quantization errors. Extensive experiments demonstrate the effectiveness of our proposed HARR. [ABSTRACT FROM AUTHOR]

INTRODUCTION: Sports video is an essential information resource. Classifying sports videos with high accuracy can effectively improve users' browsing and query effect. This project intends to study a motion video classification algorithm based on deep-learning particle filters to solve the problems of solid subjectivity and low accuracy of existing motion video classification algorithms. A critical box extraction method based on similarity is proposed. The moving video classification algorithm is studied based on a deep learning coding model. Examples of various types of sports videos are analyzed. The overall performance of the motion video classification algorithm proposed in this paper is much better than other existing motion video classification algorithms. This algorithm can significantly improve the classification performance of motion video. [ABSTRACT FROM AUTHOR]

Edge devices, due to their limited computational and storage resources, often require the use of compilers for program optimization. Therefore, ensuring the security and reliability of these compilers is of paramount importance in the emerging field of edge AI. One widely used testing method for this purpose is fuzz testing, which detects bugs by inputting random test cases into the target program. However, this process consumes significant time and resources. To improve the efficiency of compiler fuzz testing, it is common practice to utilize test case prioritization techniques. Some researchers usemachine learning to predict the code coverage of test cases, aiming to maximize the test capability for the target compiler by increasing the overall predicted coverage of the test cases. Nevertheless, these methods can only forecast the code coverage of the compiler at a specific optimization level, potentially missing many optimization-related bugs. In this paper, we introduce C-CORE (short for Clustering by Code Representation), the first framework to prioritize test cases according to their code representations, which are derived directly from the source codes. This approach avoids being limited to specific compiler states and extends to a broader range of compiler bugs. Specifically, we first train a scaled pre-trained programming language model to capture as many common features as possible from the test cases generated by a fuzzer. Using this pre-trained model, we then train two downstreammodels: one for predicting the likelihood of triggering a bug and another for identifying code representations associated with bugs. Subsequently, we cluster the test cases according to their code representations and select the highest-scoring test case fromeach cluster as the high-quality test case. This reduction in redundant testing cases leads to time savings. Comprehensive evaluation results reveal that code representations are better at distinguishing test capabilities, and C-CORE significantly enhances testing efficiency. Across four datasets, C-CORE increases the average of the percentage of faults detected (APFD) value by 0.16 to 0.31 and reduces test time by over 50% in 46% of cases. When compared to the best results fromapproaches using predicted code coverage, C-CORE improves the APFD value by 1.1% to 12.3% and achieves an overall time-saving of 159.1%. [ABSTRACT FROM AUTHOR]

How to develop students' computational thinking (CT) is an important topic faced by academics and front-line teachers. However, the solution of programming problems requires paying attention to every detail of the problem and building a solution to the problem step by step, and for beginners, they often get stuck when one of these aspects goes wrong because of the lack of metacognitive abilities. The integration of metacognitive scaffolding in project-based programming instruction can help students identify their strengths, become more aware of their learning status and identify problems in a timely manner. Therefore, this study designed a metacognitive scaffolding in four aspects: planning, monitoring, reflecting and evaluating, and assessed the effects of this scaffolding on students' CT, learning achievement and metacognitive abilities through a quasi-experimental design. The participants were 70 students aged 9–11 years in elementary school, where the experimental group (38 students) used a metacognitive scaffolding-based project-based learning approach, while the control group (32 students) used a traditional project-based learning approach. The results indicate that metacognitive scaffolding has a facilitative effect in helping students improve their CT and learning achievement, but does not significantly improve metacognitive abilities. This study provides insights into the deeper development of students' CT development and metacognitive scaffolding design. [ABSTRACT FROM AUTHOR]